Table 3.
Quantitative performance of the different methods to estimate the PRC using noisy, continuously fluctuating data.
| Method | NL = 1 | NL = 2 | NL = 3 | NL = 4 | ||||
|---|---|---|---|---|---|---|---|---|
| PC | MSE | PC | MSE | PC | MSE | PC | MSE | |
| STA | −0.089941 | 514 | −0.252694 | 562 | −0.191698 | 541 | 0.279241 | 410 |
| STEP | 0.751355 | 343 | 0.702422 | 544 | 0.671676 | 336 | 0.551901 | 387 |
| WSTA | 0.847633 | 212 | 0.776861 | 282 | 0.815145 | 318 | 0.788307 | 459 |
The WSTA performs the best on most noise levels in terms of both the MSE and Pearson correlation (PC) when compared to the directly observed PRC. An interpretation of these results is in the main text.